sbml4md: A computational platform for system-bath modeling via molecular dynamics powered by machine learning.

Area of Science:

• Computational Chemistry
• Molecular Dynamics Simulations
• Spectroscopy

Background:

• Simulating nonlinear vibrational spectra of molecular liquids is computationally challenging.
• Existing methods often require empirical fitting, limiting accuracy and applicability.
• Capturing vibrational anharmonicity and intermolecular couplings is crucial for realistic models.

Purpose of the Study:

• Introduce sbml4md, a novel algorithm and software package.
• Enable accurate simulation of nonlinear vibrational spectra using molecular dynamics (MD) trajectories.
• Provide parameters for the Hierarchical Equations of Motion (HEOM) framework.

Main Methods:

• Leverage machine learning (ML) techniques to extract model parameters.
• Account for vibrational anharmonicity, intermolecular couplings, and bath correlation functions.
• Integrate classical MD approaches with HEOM for enhanced dynamical modeling.

Main Results:

• sbml4md obviates empirical fitting by directly extracting parameters from MD data.
• The software enables modeling of heterogeneous environments with spatial and temporal variations.
• Enhanced optimization efficiency by including intermolecular vibrational contributions.

Conclusions:

• sbml4md provides a flexible and scalable framework for simulating linear and nonlinear spectra.
• The approach minimizes empirical input for realistic simulations.
• Facilitates numerically "exact" simulations of nonlinear vibrational spectra within the HEOM framework.